Hacker News Reader

Comment by sudosysgen | Hacker News Reader

The big issue with this design is that it would be necessary to change the curvature as the focal length changes. If they can get that working reliably, it would be an absolute game changer - lenses would get much cheaper, much lighter, and of higher quality too.

View on HN

Lenses never get cheaper. Look at micro four thirds. At the high end they don’t get lighter except for really long wide lenses. That’s because they got heavier when autofocus was added. The original manual focus Nikon 500mm f4 “only” weighs 3kg. It’s lighter than the current $12k model which shed about 2.4kg since the first autofocus version.

Plastic is easiest way to make a lens lighter and high end buyers are adverse to plastic. People buy lenses by the kilogram so to speak.

That is true. That being said, there is a renaissance in third party lenses, and their quality as well as their autofocus performance is increasing, fast.

If you are willing to buy third party, maybe even Chinese lenses, they are getting cheaper. If we ever get a true variable curvature sensor, designing lenses will get so much easier and cheaper that I can see even more third party lenses of even higher quality, at an even *lower price.

In any case, I agree with the point about plastic lenses.

That being said, there is a renaissance in third party lenses, and their quality as well as their autofocus performance is increasing, fast.

While I agree with you, the third-party lenses that have gotten notably better are Sigma’s, which are no longer cheaper, and are also just big ol’ chonkers now.

https://www.dpreview.com/products/Sigma/lenses/sigma_85_1p4_...

Ah, I'm not really thinking of Sigma, I'm talking of Neewer, Viltrox, Mitakon, and in some respects Tamron, and so on. They are making very good primes nowadays, and Viltrox even makes good AF primes now.

What's missing is even more AF and zoom.

Third party medium format system lenses are not something I see when browsing. Though I would not be stunned to come across a Vmount product. I would attribute most of it to market size and adapters for the mirrorless systems and maybe the availability of used 645 lenses.

35mm is a whole different scale of market with entirely different buyers. The entry point cost to medium format is more than the 99th percentile of 35mm like systems.

I now see the Fuji G series has third party lenses.

I don't mind the plastic lenses I use on my D5600, but I know they wouldn't survive bumping into something. People who do it professionally bump into stuff and won't always be where they can replace it, so they need sturdier material.

The bumper on your car is probably plastic. It will survive 1500 kg moving at 8kmh. If you bang a lens, the most likely breakage will be the precision optical system going out of whack. By the time the housing cracks, that’s likely to have already happened as well.

Even then it might take a bit before you even notice or -care-.

My best Sony E 18-55 is ever so slightly cocked out of alignment when at it's past ~45mm or so, yet you could never tell from the photos.

Yup. The cool thing about phase detect autofocus is that it will correct for all first-order aberrations, such as tilt in the lens, at the focus point.

My old D3100 and its kit, plastic 18-55 lens have been through hell and back. It's knocked into more things than I can count hanging over my shoulder or around my torso. I recently sold the system to a friend and had the chance to test it out, it still works great

I would say Nikon lenses are pretty sturdy

I didn't really mean that the lens barrel would be plastic - I think that a healthy mix of fiber reinforced plastics and metals is the best - I was also thinking of optical elements themselves being plastic, which allows for much cheaper prices.

That being said, most lens damage is in the filter thread or in optical alignment, metal vs plastic doesn't make that big of a difference.

My m43 lenses are cheaper than my E Mount lenses. 12-40 2.8 Olympus lens is less than half the price of a Sony lens with the equivalent focal length. Most my e Mount lenses are over $1k and all but one of my m43s lenses are under $1k. Glass cost more for the larger sensor.

Your Olympus 12-40 is slower everywhere but in the teleend but has a larger zoom range than my Sony FE 28-70 f/3.5-5.6, however my lens comes at a third of the cost.

So, for essentially equivalent glass, the M4/3 lens is more expensive than full-frame.

For the same price as your 12-40, there is the Tamron 28-75 f/2.8 for Sony FE, which captures four times as much light.

So no, glass really doesn't cost more for the larger sensor. Equivalent glass is often cheaper on Full-frame, which is why I went FF as I had a limited budget.

This is the reason why so many people are buying speedboosters, because FF lenses are often just better value.

Constant 2.8 is faster than the variable aperture lens you posted. It is equivalent to the GMaster 24-70 or the canon L 24-70 at less than half the price. Its not on the same level as a Tamron, it’s made of machined aluminum and feels solid. Tamron is made out of plastic.

No, it isn't. Constant 2.8 at M43 is exactly equal to a constant f/5.6 on FF. Indeed, at the FoV equal to 50mm on FF, you would need 25mm on M43. At f/5.6, the aperture diameter of the 50mm FF lens is 50/5.6=8.93, and the M43 lens would have an aperture diameter of 25/2.8=8.93mm.

This means that the lenses are equivalent. Indeed, the amount of light captured by a lens from a point source is proportional to the angle of the cone with its tip at the point and it's disk at the aperture. And so, since the FoV is the same, you will be at the same distance from the point for the same image, and since the diameter of the aperture is the same, the two cones will be identical, and thus the amount of light will be the same.

For depth of field, the calculation is similar, and they are indeed also equal.

And so, an f/5.6 aperture on FF is equal to an f/2.8 aperture on an M43 lens of the same FoV.

As a conclusion, your 12-40mm f/2. 8 lens is exactly equivalent to a 24-80 f/5.6 lens on FF. And so, a 28-75 f/3.5-5.6 lens on FF is faster in average but has a slightly smaller zoom range. In the end, the variable aperture lens is optically superior, at a third of the price.

As for the Tamron, it is indeed made of plastic, but I like this because it makes it lighter and barely less resistant as the G-Master. It is still capable of gathering four times more light.

If you want to actually test this, we can print two test charts and put them one meter from the camera then one meter from each other in a line, and you can take a picture of them at 25mm f/2.8 on your Olympus, I'll take the same picture at 50mm f/5.6 on my A7ii, and you'll see that the depth of field will be exactly the same. If we had a way of controlling lighting, we would see that at the same shutter speed the noise levels of the two images will be very similar too.

DOF is different than speed of the lens. F value for aperture is standard no matter the size of sensor and sensor will get the same amount of light if the shutter speed are the same.

I explained to you exactly why this is incorrect. The amount of light the sensor catches from a point source depends on the diameter of the aperture and the distance from the point. So, at the same FoV, the diameter of the aperture is what matters.

The light intensity at the same f number is indeed the case, however, since the image is smaller, then amount of light is four times smaller for an M4/3 lens at the same FoV. Another way to understand this is to realize that the focal length is smaller, therefore the image for the same object is smaller, therefore despite an equal light intensity the total light is the same.

DoF is not actually any different from lens speed. There is a physically indestructible link where increasing the amount of light from the same object reduce the depth of field. DoF and amount of light are two faces of the same coin.

I highly suggest actually doing the math. It's a very counter-intuitive concept. Take a point source of light, take some distance, then calculate the angle of the triangle formed by the point and the opposite points on the aperture. This angle is directly proportional to the light. You will see that the angle is equal in both.

Then, to explain why the ISO is going to be four times as high in the FF camera, calculate the intensity of the image by taking the source to now be a sphere, and calculating the image size using the thin lens equation. You will se that the image will be twice the size with the 50mm FF lens, and thus the light intensity will be half that, which is why ISO is going to be four times higher in the FF camera.

This is also why you can get away with ISOs four times on full-frame than on M4/3.

Again, if you don't believe me, you really have to do the math in your own, failing that you should do an experiment and look at the noise levels at the same shutter speed at f/2.8 vs f/5.6 in the same speed with equivalent focal lengths. You will see.

Parent (sudosysgen) is correct, I don't know why they're being downvoted.

Another way to look at it is, if you have two setups with different sensor sizes, but the same relative aperture (i.e. f number), same ISO, same sensor resolution, and equivalent focal lengths (i.e. scaled by crop factor), you will have more noise in the image from the smaller sensor. As the parent stated, this is purely due to physics: the lens in the setup with the smaller sensor has a smaller absolute aperture; thus less total light is entering the camera.

Less luminous flux arrives to the surface of the smaller sensor. However the illuminance is the same due to the photons being spread over less area.

Measuring lux at each sensor would provide similar readings.

You get the same exposure, not the same amount of light. A larger sensor requires more total light for the same exposure (the light is spread out over more area, after all). This is true equally of film and digital.

Lenses absolutely get cheaper if you look at the lenses that are actually bought and sold in any number. The number of people using a 500mm f/4 is vanishingly small, and wondering about their price is a bit like asking why Ferraris still cost a lot.

For comparison, let's look at the most popular lens in the world, the 50mm. I'll focus on Canon's f/1.8 series of the nifty-fifty, just to keep consistent.

1975: $79.95 ($386.95 today)

1978: $67.95 ($271.37 today)

1986: $102 ($242.33 today)

1992: $44.95 ($83.42 today)

2014: $125 ($137.49 today)

Today: $94.00 (discontinued) or $125 (new model)

While the list price of the 50mm tends to sit in the $50-125 range, the inflation adjusted cost of this lens has dropped fairly consistently over the years. And remember that this is covering a pretty massive range of lens technology; a modern 50mm f/1.8 is lighter, focuses closer (0.35m vs 0.6m), has an extra diaphram blade in it (7 vs 6), contains several multi-coated glass elements, and has an auto focus motor built into the body of the lens. The value per dollar of a modern photo lens is genuinely massive compared to what was available in the past.

Well, the low point seems to be 1992 at $88. Which sort of matches my experience.

I think it's misleading to just look at Canon SLRs. SLRs are dying. Looking at the most modern mounts from all companies:

* Canon 50mm f/1.8 R-mount: $200

* Fuji 35mm f/2: $200

* Nikon 50mm $-mount f/1.8: on sale for $500 (normally $600)

* Olympus 25mm f/1.8: $300

* Panasonic 25mm f/1.8 MFT: On sale for $150 (normally $250)

* Panasonic 50mm f/1.4 L-mount: $2300 (although Sigma makes one for just $430).

* Sony 50mm f/1.8 E-mount: $250

I've adjusted to 50mm for crop factor.

I think the Canon f/1.8 50mm is sort of an outlier for low-cost in 2020. On the whole, prices have been creeping up (as has sharpness, AF speed, flare control, etc.). I feel like lenses are as expensive as they've been in decades.

These are also all new mounts, while the SLR mounts are decades old. If there's anything we would expect to reset the cost of equipment, it's a new and incompatible mount which saps the used market of its entire supply and lets manufacturers to charge higher rates for new and shiny gear. Also, inflation adjusted a lot of those are still cheaper than the original 1971 lens, despite having more features.

By the by, I really don't get the rush to mirrorless. I hate electronic viewfinders, the really small bodies are hard to handle with bigger lenses, and the prices are outrageous. I'll probably keep using SLRs for the next 20-30 years.

Not to nitpick too much, but the first two examples of lens prices are (presumably) for FD mount. The third most likely as well because the EOS mount was introduced in 1987 (per a quick wikipedia). The early versions would have been breach mount, while the 1986 would have been the later bayonet style (introduced in 1981).

Anyway, my on-the-spectrum reaction aside, I started exploring mirrorless because it facilitated making different types of pictures. Part of that facilitation was it's low cost. A used NEX-3, a fotodiox shift adapter (for K mount), and some old manual focus primes was all in for about $300. I got a camera with front shift that fit in a (largish) pocket. 14mp will about cover 10"x15" printed at 300dpi and is overkill for anything online.

I find manual focus with focus peaking satisfying. It's slower but when it misses it misses because of my technique not the way the camera interprets the scene. I don't get as many mis-focused pictures. I learn from those I get. The fixes are obvious. I don't feel I need to make more "just to be safe."

That's the biggest thing for me with mirrorless. My workflow has become much simpler. I don't take extra pictures just to be sure. I can see how the shadows and highlights will render in the viewfinder. I can adjust the contrast and saturation and color profile and white balance in camera while looking at the scene.

But, it's made me heretical. I no longer shoot RAW...though there are situations where I would...just JPG. My workflow is simpler because I don't spend time trying to fix pictures. Color, shadow, and highlights were either pretty close in camera or I didn't take it.

Even better for my workflow, JPG's are tiny. Memory cards don't fill up. Disks don't fill up. Downloads are faster. Backups are faster. And there are fewer files because I know that I got it when I made it.

But...I don't shoot sports anymore. I don't shoot wildlife anymore either. I'm not chasing birds in flight. My pictures were shifting as my gear was. Both still are. Mirrorless changed the way I was making pictures. The more a piece of gear does that, the more interested in it I am. YMMV.

I recommend RAW+JPEG.

I hate editing RAW files, but in 30 years, I'll still have them and might have a use for them. In 300 years, my ancestors will have them. Seems like image processing algorithms might advance by then.

I once reran my photos from the early '00s through a modern raw processor, and the difference was night-and-day. No tweaking, no nothing. Just modern versus ancient noise reduction, color algorithms, etc.

Disk space is cheap enough.

I use the same non-destructive editor, Darktable, on my JPG originals as for RAW. Make non-destructive edits and export. My workflow is the same though JPG requires a less pixel pipeline during post processing.

To put it another way the only difference is that my “archived negative” is a JPG rather than a RAW. But in 300 years I would only bet, and not very much, on physical prints. I’d give JPG a reasonable chance thirty years from now. Proprietary RAW files will probably be readily accessible in fifteen years for the most part.

Most of what JPG does is free me from a “raw+jpg just in case.” I don’t think defensively like that anymore...I don’t burden myself with a lot of “a serious photographer would.” What I am making is a picture. There are tradeoffs and the actual pictures I make (not the ones I might possibly make) are how I think about cost and benefit.

Disk space is cheap and so I use it write only. The time I spend on file management is expensive, so again write only. RAW + JPG requires much more file management than either alone. Should the JPG and RAW receive the same number of stars? The same tags? Always? I don’t want to spend time thinking about that.

For me, RAW represents an obligation to future work. The picture is never done when it is made. Not ready upon download.

JPG frees up energy for making more pictures. YMMV.

For the past few years, an older OM-D mirrorless was by far the best, cheapest camera on the market. New-old-stock or used.

I don’t get that. The existing flat sensors don’t change curvature as focal length changes?

That's because current lenses correct field curvature so that the image plane is flat. This takes extra glass and can degrade image quality.

If you could have a sensor with the correct curvature for the focal length of the lens, you don't need to correct field curvature anymore, which means that you need less glass and have higher quality.

If you have the wrong curvature for the focal length, some field correction will still be required, which defeats the advantage.

In the real world, it's even more than just focal length that changes the required curvature.

Will it defeat the advantage or lessen it? For example, some people still need glasses after LASIK but still opt for it since they get end up with needing weaker lenses.

No, absolutely. The sensor is flat and when you change the focal length, the sensor gets the same “light/illumination”. That is the plus of a curved sensor: better image/photo at the border of the frame because there is always the same distance between the lens and the sensor.

Maybe they’re developing a fixed lens camera like the Leica Q or the old (defunct) R1XR.

Maybe! A fixed lens médium format caméra would be very... Interesting to the very least.

It was not unheard-of back in the film days. The cameras, like the Fuji GW690, were not exactly small, but you could carry them around on a neck-strap and use them without a tripod.

1. http://camera-wiki.org/wiki/Fujica_GW690

True, but the inherent costs of a medium format sensor (not even taking into account the likely extra costs of a curved sensor) mean that the bare minimum you could sell this fixed-lens camera for would be several thousand dollars. I just don't see a lot of photographers being willing to drop that much money on a camera that they can never expand the capabilities of.

The Leica Q2 sells for $5000 dollars - I'm guessing it's (or was until Covid hit) selling decently. There would be many who'd prefer a light medium format with fixed lens instead of the Q2 if it's under 7K or 8K. It'll blow the Q2 out of the water.

Leica Q is one of the best selling camera for Leica, infact they raised the price 3/4 years ago. Also the Q2 had a 200€ price increase in EU (https://www.macfilos.com/2019/08/01/leica-q2-price-rises-in-...).

Leica has already a medium format camera (S) with interchangeable lenses, its super expansive but a medium format Leica with fixed lens for (example) 9000$ I think would sell really a lot.

Also a Sony medium format for 7/8000€ would sell a lot (more than Leica surely). We’ll see...

Quite a lot of medium format cameras were fixed lens. Pretty much all the TLR’s (except some Mamiya’s). All the folders (to a first approximation), the Brownies, and even today all those Holgas.

A GS645w isn’t much larger than a modern 35mm digital, and is certainly lighter.

Is it defunct? It's still listed on their website.

https://www.sony.co.uk/electronics/cyber-shot-compact-camera...

Maybe Sony is still salling it but was coming out 4/5 years ago without updates. I don’t know who want to buy it instead an A7

It still has unparalleled small size. The a7C is a nice alternative but it's not 42 megapixels. And there are not many compact 35mm lenses with good image quality, and the compact ones tend to be f/2.8 rather than f/2.

Reply on news.ycombinator.com